Sage Journals: Discover world-class research

Abstract

As an infectious disease, malaria consumes around 250 million yearly clinical cases and with more than half a million annual deaths. It has shown tremendous burden for the economic and social life of many countries around the world, particularly in the tropical and developing nations. The conventional wisdom claims that the prevalence of malaria infection either prolongs or should be positively correlated with outbreaks of civil conflicts. We contend that malaria infection should deter civil conflict occurrences because warming parties should avoid engaging each other in areas with rampant malaria infection. We test the hypothesis with 20 years of geo-referenced panel data of conflict event and malaria risk from Sub-Sahara Africa. Our result renders strong support for our hypothesis that areas with more malaria infection tends to have less civil conflicts.

Keywords

Civil war infectious disease malaria Sub-Sahara Africa

Introduction

Malaria, a prevalent tropical disease, significantly impacts global human settlement. Accounting for approximately 250 million annual clinical cases and over half a million deaths, malaria imposes substantial economic and societal burdens, especially in developing nations within the tropics (Sachs and Malaney, 2002: 680). Malaria infection has been identified as a major barrier to economic development in the Global South (Worrall et al., 2005).

A parasitic infection transmitted via the female Anopheles mosquito, malaria profoundly affects human health and military efficacy in tropical climates (Holt et al., 2002). The disease’s prevalence in tropical forests and jungled marshes, common mosquito habitats, has made it a most daunting challenge for warfare in these regions (Slater, 2009). History is replete with instances where malaria impeded effective warfighting in the tropics, particularly against unacclimated troops. It notably hindered military efficiency during World War I, the Pacific campaigns of World War II, and American engagements in Indochina during the Cold War (Beadle and Hoffman, 1993; Bello, 2005; Brabin, 2014).

Although malaria’s impact on international conflicts is well recognized, its systematic examination in the scholarship of civil conflict dynamics remains rare. A handful of existing studies concur on a positive correlation between malaria prevalence and civil conflicts. Yet they diverge on whether pathogen prevalence predisposes countries to civil war or if civil war augments disease incidence (Hendrix and Gleditsch, 2012; Letendre et al., 2010). Furthermore, Bagozzi (2016) argues that malaria prevalence may extend the duration of civil conflicts.

In this article, we hypothesize that malaria prevalence can lead to reduced civil conflict incidences. This theory arises from the assumption that both government and rebel forces would likely strive to evade areas stricken with the disease, resulting in fewer militarized civil conflicts in regions prone to severe malaria infection. To empirically test this hypothesis, we diverge from the traditional methodology in prior research, which uses country as the unit of analysis. Instead, we delve into the influence of malaria prevalence on the actual battlefield selection at the subnational level.

We employ granular data from the Uppsala Conflict Data Program Georeferenced Event Dataset (UCDP-GED) (Sundberg and Melander, 2013) and GIS-referenced malaria data from the Malaria Atlas Project (MAP) (Guerra et al., 2007; Hay and Snow, 2006). We focus on Sub-Saharan Africa, a region characterized by both high malaria prevalence and frequent civil conflicts (Buhaug and Rød, 2006). Our empirical analysis robustly supports our hypothesis, revealing a negative correlation between malaria prevalence and the occurrence of civil conflict battles.

This article is organized as follows. First, we situate our study in existing literature and explain our theory about how malaria could decrease civil conflict occurrences. Next, we introduce our data sources, empirical models, and statistical results. Using South Sudan as a case study, we further illustrate the relationship between malaria infection and civil conflict dynamics. We conclude the article by noting the limitations of our study and suggesting directions for future research on pandemic and civil conflict dynamics.

Malaria and civil conflicts in Sub-Sahara Africa

Malaria’s global infection patterns are different across continents because of its association with the prevalence of certain mosquitos. Of these, Sub-Saharan Africa has the most stable rates of malaria infection (Carter and Mendis, 2002: 567). Meanwhile, Sub-Saharan Africa has received considerable scholarly attention within the literature on civil conflicts, given its large number of nation-states within the African continent and its expansive geographical span. We argue that malaria has a temporal-spatial effect on geographical locations of civil conflicts in Sub-Saharan Africa.

Bagozzi argues that malaria should be considered a rough terrain factor that can “asymmetrically enhance rebel group’s defensive capabilities and thus prolong the duration of civil war” (Bagozzi, 2016: 817). He points out that the cost of malaria will be higher for larger government troops, who are less likely to develop immunity to malaria than rebel groups due to frequency of troop rotation. Meanwhile, the presence of malaria can prevent government from constructing crucial infrastructure and poses challenges to the projection of government forces (Bagozzi, 2016: 817–8).

Does this effect on prolonging civil war mean there are more frequent battles among warring parties? Historical evidence suggests the opposite: malaria has been recognized as a significant factor that hinders effective fighting in tropical areas. While malaria’s rough terrain effect may favor the rebel forces, it is also reasonable to expect that government forces would avoid engaging with rebel forces in places with high infection rate of malaria. Such engagement would diminish the advantage enjoyed by the government troops while favoring the rebel forces. Consequently, government forces would naturally seek to fight the rebel forces in other locations where they have a better chance of success. Additionally, while rebel forces may develop some immunity to malaria, this immunity is not necessary long-lasting (Struik and Riley, 2004). Repeated malaria infection can still have negative health consequences (Hafalla et al., 2011; Weiss et al., 2010). Therefore, neither government nor rebel forces would willingly engage in military battles in area where there is prevalent malaria risk. Moreover, escalated fighting between government and rebel forces often leads to the displacement of civilian population seeking refuge in more challenging terrains, thereby increasing their vulnerability to malaria. As a result, we expect malaria infection to deter militarized battles, leading us to hypothesize a negative correlation between battle location and malaria infection.

Data

To examine the local-level relationship between conflict and malaria risk, we compile a panel dataset spanning 20 years (2000–19) in Sub-Saharan Africa. The dataset contains geo-referenced information on both conflict occurrences and malaria risk. Our dependent variable is the occurrence of civil conflicts and is constructed using UCDP-GED. We specifically focus on the subset of state-based violence, which includes events where at least one side is the state. We also construct an alternative dependent variable using records of battles from the Armed Conflict Location and Event Data Project (ACLED) for robustness checks (Raleigh et al., 2010).

Our independent variable is the risk of malaria infection, which we measure using the proportion of children 2–10 years of age showing detectable Plasmodium falciparum parasite (Pfpr 2-10).¹ We obtain these data from MAP, a public health data initiative known for its comprehensive cross-national geo-referenced malaria risk data. MAP employs a two-step meta-analysis process to construct the malaria risk measures (Bhatt et al., 2015; Pfeffer et al., 2018). First, they aggregate geo-referenced surveys on mosquitos and falciparum parasite prevalence from various sources. Second, they fit spatial-temporal epidemiological models using the vector data to interpolate malaria risk data in area-years where direct measurements are not available.

To account for potential confounders, we include a set of covariates in our statistical models. Malaria risks can be associated with state capacity, roughness of physical terrain, level of development, and population mobility, which are widely recognized determinants of civil conflicts (Chang and Wei, 2019; Datta and Reimer, 2013; Martens and Hall, 2000). To account for confounders, we control for indicators of these factors in our statistical models. First, we control for indicators of physical rough terrain, including forest area, mountainous area, and precipitation level. Second, we control for indicators of development level, including night-time light, urban area, irrigated area, and population. Thirdly, we control for state capacity measured by distances to border and distance to capital (Chen and Nordhaus, 2011). Finally, we account for population mobility by controlling for the presence of refugee camps.

We obtain most of the above covariates from the PRIO-GRID 2.0 database (Tollefsen et al., 2012). To complement these, we utilize data from four additional sources: nighttime light data from new data work that harmonizes the pre- and post-2013 satellite data (Li et al., 2020), forest coverage data from the Global Forest Change data (Hansen et al., 2013), precipitation data from The Tropical Rainfall Measuring Mission (Huffman et al., 2007), and location for refugee camps from Geo-Refugee (Fisk, 2017).

All the data are transformed into PRIO-GRID vector grid networks with resolutions of 0.5 × 0.5 decimal degrees.² By merging these datasets, we construct a 20-year (2000–2019) panel of raster data covering grids located in Sub-Saharan Africa (by United Nation’s definition).³

In Figure 1, we present a visual depiction of our dependent and independent variables, along with their correlation. Panel (a) and (b) show the spatial distribution of state-based conflicts and malaria risk in Sub-Saharan Africa, showcasing significant variation across the region. Panel (c) depicts the bivariate correlation between state-based conflicts and malaria through a boxplot by group. It shows that grid-years with recorded state-based conflicts (top box) on average exhibit lower malaria risk than those without conflicts (bottom box), providing initial support for of our hypothesis.

Figure 1.

Spatial distribution of state-based conflicts and malaria risk.

Empirical models and results

We fit logistic regression models to systematically investigate the relationship between malaria risk and the occurrence of conflict. Our dependent variable is a binary indicator of conflict occurrence at the grid-year level. Malaria risk is our main independent variable. We also control for a set of conventionally important predictor of civil conflicts, including physical rough terrain, development level, state capacity, and human mobility. To address spatial-temporal dependencies, we incorporate spatially and temporally lagged dependent variables—the former is defined as the proportion of a grid’s 10 neighboring grids that have conflicts in the previous year and the latter whether a grided cell has conflict occurrence in the previous year (Ward and Gleditsch, 2008).⁴ Formally, for Grid i in Country j in Year t

\Pr (C o n f l i c t_{i, j, t} = 1) = L o g i t (β_{0} + β_{1} C o n f l i c t_{i, j, t - 1} + β_{2} C o n f l i c t_{n e i g h b o r, t - 1} + β_{3} M a l a r i a_{i, j, t - 1} + β_{4} R o u g h_{i, j, t - 1} + β_{5} D e v_{i, j, t - 1} + β_{6} S t a t e C a p_{i, j, t - 1} + β_{7} H u m a n M o b i l i t y_{i, j, t - 1} + β_{7} C o u n t r y_{j} + β_{8} Y e a r_{t})

Our results suggest that malaria risk has a statistically and substantially significant negative effect on the odds of conflict occurrence. Table 1 highlights the results from our statistical analysis. Specifically, Model 1, our baseline model, includes only two autoregressive terms. Model two to six include malaria risk as our independent variable of interest, alongside various combinations of control variables. The models consistently indicate that malaria risk retains its negative association with conflict occurrences, even when accounting for physical rough terrain, development level, state capacity, and human mobility.

Table 1.

Malaria risk (Pfpr2-10) is a statistically and substantially significant predictor of state-based conflict occurrence.

	(1)	(2)	(3)	(4)	(5)	(6)
Malaria risk (Pfpr2-10)		−0.21^*** (0.04)	−0.21^*** (0.05)	−0.18^*** (0.05)	−0.18^*** (0.05)	−0.19^*** (0.05)
Control variables
Temporally and spatially lagged y	✓	✓	✓	✓	✓	✓
Rough terrain			✓	✓	✓	✓
Development				✓	✓	✓
State capacity					✓	✓
Human mobility						✓
Fixed effect	Country-year two-way fixed effect
AIC	14,692.28	14,669.84	11,310.50	11,224.78	11,185.40	11,151.77
BIC	15,255.65	15,242.72	11,892.36	11,844.01	11,823.39	11,799.14
Log likelihood	−7287.14	−7274.92	−5593.25	−5546.39	−5524.70	−5506.88
Deviance	14,574.28	14,549.84	11,186.50	11,092.78	11,049.40	11,013.77
Num. obs	103,645	103,567	87,977	87,746	87,746	87,746

^***p < 0.01.

^**p < 0.05.

^*p < 0.1 Full results are reported in Table A1of the Online Appendix.

Across models, the estimated coefficients for malaria risk have larger absolute values compared to all those of control variables. This suggests that the magnitude of the effect of malaria is the largest even comparing with conventional important predictors of conflict occurrences.⁵ Figure 2 shows the simulated odds ratios of malaria risk and control variables. The estimated odds ratio of malaria risk ranges from 0.81 to 0.84 across the models. This means for every 1 standard deviation increase in the proportion of children aged 2–10 with Plasmodium falciparum in the current year, the odds of conflict occurrence in the next year decrease by 16%–19%. Notably, the effect of malaria risk on conflict odds surpasses that of conventionally important predictors of conflict occurrence in our models: roughness of physical terrain (forest area, mountainous area, and precipitation level, development level (night-time light, urban area, and population), and state capacity (distance to border and capital)—none of them shifts the odds of conflict by more than 16%.

Figure 2.

The odds ratio of malaria risk is the largest compared to other covariates.

Our findings withstand robustness checks involving alternative measures and model specifications. We fit models with only spatially lagged dependent variables. Considering the rare occurrence of conflicts (only 3% of the grid-year report state-based civil conflicts), we fit rare event logistic regressions (King and Zeng, 2001).⁶ To address possible measurement error associated with our malaria risk indicator, we fit models with an alternative independent variable: the number of newly diagnosed P. falciparum cases per 1000 population. Additionally, to account for possible measurement error of the UCDP dataset, we fit models with the dependent variable measured by the ACLED dataset. Across all these robustness checks, malaria risk consistently emerges as a statistically significant predictor, leading to a reduction in conflict odds.⁷

South Sudan as an illustrative case

The case of South Sudan serves as a compelling example of the negative association between malaria risk and militarized conflicts in specific regions. Since gaining independence in 2011, South Sudan has exhibited spatial variation in both conflicts and malaria, as depicted in Panels (a) and (b) of Figure 3. To further explore this relationship, we focus on the subset of the grids in South Sudan and fit them to the statistical models shown in Table 1. Our results, presented in Panel (c) of Figure 3, indicate that malaria risk is a statistically significant predictor of reduced conflict odds. Moreover, the magnitude of the effect is large, with a standard deviation increase in malaria risk associated with 13%–39% reduction of conflict odds.⁸

Figure 3.

Malaria risk decreases conflict odds in South Sudan from 2010 to 2019.

Qualitative evidence further supports our statistical analysis of the case of South Sudan. The country’s struggle for independence from Sudan and subsequent civil war have resulted in prolonged violence, causing significant damage to physical infrastructure, social institutions, and the healthcare system. Malaria remains the most prevalent cause of morbidity and mortality in South Sudan (Idris et al., 2022). Specifically, it accounts for 20%–40% of morbidity cases, with over 20% of reported deaths occurring in health facilities, and 30% of hospital admissions related to the disease (Pasquale et al., 2013). During the recent civil war in South Sudan, the confrontation between rebel and government forces led to the displacement of tens of thousands of individuals, forcing them to seek refuge in hostile, non-conducive environments such as the bush. In these areas, they lacked access to essential medical facilities and mosquito nets. According to Medecins Sans Frontieres (2021), this led to increased risk of malaria infection. While it is difficult to provide definitive evidence of the avoidance of malaria-infested areas by government or rebel forces, existing reports suggest that civilians who sought refuge in these areas were at the highest risk of infection.

Conclusion and discussion

In this article, we have extensively examined the relationship between malaria infection distribution and civil conflicts in Sub-Saharan Africa using granular geo-referenced data. Our findings robustly support our hypothesis of a negative correlation between the two.

It is worth noting that our results do not contradict previous study arguing that malaria can prolong civil conflict durations (Bagozzi, 2016). While malaria prevalence may reduce the frequency of militarized battles, it does not exclude the possibility of prolonged civil wars. In another word, in areas with rampant malaria, civil wars may be “colder”—they last longer but exhibit a lower intensity.

This study contributes to the understanding of how infectious diseases, using malaria as an example, can affect civil conflict dynamics in Sub-Saharan Africa. Although previous studies on civil conflicts and infectious diseases have suggested a positive correlation between the two (Hendrix and Gleditsch, 2012; Letendre et al., 2010), our results suggest a more nuanced relationship.

This study has clear limitations. Particularly, the potential bias in news reporting—a common issue in studies relying on civil conflict news reports—could affect the accuracy of our findings. For example, journalists’ reluctance to travel to malaria-prone regions could results in underreporting of conflicts. Future research can consider these reporting biases and further investigate the impact of different types of diseases on civil conflicts and vice versa.

Supplemental Material

Supplemental Material - Infectious disease and political violence: Evidence From malaria and civil conflicts in sub-saharan Africa

Supplemental Material for Infectious disease and political violence: Evidence From malaria and civil conflicts in Sub-Saharan Africa Haohan Chen, Zifeng Wang, and Enze Han in Research & Politics.

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Publication made possible in part by support from the HKU Libraries Open Access Author Fund sponsored by the HKU Libraries.

Correction (June 2025):

The article has been updated with correct dataverse link in the supplementary material section. For more details, please see the correction notice: .

ORCID iD

Enze Han

Supplemental Material

Supplemental material for this article is available online.

The files can be found at

Notes

References

Bagozzi

(2016) On malaria and the duration of civil war. The Journal of Conflict Resolution 60(5): 813–839.

Beadle

Hoffman

(1993) History of malaria in the united states naval forces at war: world war i through the vietnam conflict. Clinical Infectious Diseases 16(2): 320–329.

Bello

(2005) To go where no han could go for long: malaria and the qing construction of ethnic administrative space in frontier yunnan. Modern China 31(3): 283–317.

Bhatt

Weiss

Cameron

, et al. (2015) The effect of malaria control on plasmodium falciparum in africa between 2000 and 2015. Nature 526(7572): 207–211.

Brabin

(2014) Malaria’s contribution to world war one – the unexpected adversary. Malaria Journal 13(1): 497.

Buhaug

Rød

(2006) Local determinants of african civil wars, 1970–2001. Political Geography 25(3): 315–335.

Carter

Mendis

(2002) Evolutionary and historical aspects of the burden of malaria. Clinical Microbiology Reviews 15(4): 564–594.

Chang

W-Y

Wei

(2019) Natural resources and infectious diseases: the case of malaria, 2000–2014. The Social Science Journal 56(3): 324–336.

Chen

Nordhaus

(2011) Using luminosity data as a proxy for economic statistics. Proceedings of the National Academy of Sciences of the United States of America. DC, USA: National Academy of Sciences.

10.

Datta

Reimer

(2013) Malaria and economic development. Review of Development Economics 17(1): 1–15.

11.

Fisk

(2017) Geo-refugee: a refugee location dataset. Cambridge, MA: Harvard University.

12.

Guerra

Hay

Lucioparedes

, et al. (2007) Assembling a global database of malaria parasite prevalence for the malaria atlas project. Malaria Journal 6(17): 1–13.

13.

Hafalla

Silvie

Matuschewski

(2011) Cell Biology and immunology of malaria. Immunological Reviews 240(1): 297–316.

14.

Hansen

Potapov

Moore

, et al. (2013) High-resolution global maps of 21st-century forest cover change. Science 342(6160): 850–853.

15.

Hay

Snow

(2006) The malaria atlas project: developing global maps of malaria risk. PLOS Medicine 3(12): 2204–2208.

16.

Hendrix

Gleditsch

(2012) Civil war: is it all about disease and xenophobia? a comment on letendre, fincher and thornhill. Biological Reviews 87(1): 163–167.

17.

Holt

Subramanian

Halpern

, et al. (2002) The genome sequence of the malaria mosquito anopheles gambiae. Science 298(5591): 129–149.

18.

Huffman

Bolvin

Nelkin

, et al. (2007) The trmm multisatellite precipitation analysis (tmpa): quasi-global, multiyear, combined-sensor precipitation estimates at fine scales. Journal of Hydrometeorology 8(1): 38–55.

19.

Idris

Ayeni

Iyamu

, et al. (2022) Factors influencing severity of recurrent malaria in a conflict-affected state of south sudan: an unmatched case-control study. Conflict and Health 16(34): 1–10.

20.

King

Zeng

(2001) Logistic regression in rare events data. Political Analysis 9(2): 137–163.

21.

Letendre

Fincher

Thornhill

(2010) Does infectious disease cause global variation in the frequency of intrastate armed conflict and civil war? Biological Reviews 85(3): 669–683.

22.

Zhou

Zhao

, et al. (2020) A harmonized global nighttime light dataset 1992–2018. Scientific Data 7(168): 1–9.

23.

Martens

Hall

(2000) Malaria on the move: human population movement and malaria transmission. Emerging Infectious Diseases 6(2): 103–109.

24.

Medecins Sans Frontieres (2021) South sudan at 10: an msf record of the consequences of violence. Geveva, Switzerland: Medecins Sans Frontieres.

25.

Pasquale

Jarvese

Julla

, et al. (2013) Malaria control in south sudan, 2006–2013: strategies, progress and challenges. Malaria Journal 12(374): 1–13.

26.

Pfeffer

Lucas

TCD

May

, et al. (2018) malariaAtlas: an R interface to global malariometric data hosted by the malaria atlas project. Malaria Journal 17(352): 1–10.

27.

Raleigh

Linke

Hegre

, et al. (2010) Introducing ACLED: an armed conflict location and event dataset. Journal of Peace Research 47(5): 651–660.

28.

Sachs

Malaney

(2002) The economic and social burden of malaria. Nature 415(6872): 680–685.

29.

Slater

(2009) War and disease: biomedical research on malaria in the twentieth century. New Brunswick, NJ: Rutgers University Press.

30.

Struik

Riley

(2004) Does malaria suffer from lack of memory? Immunological Reviews 201(1): 268–290.

31.

Sundberg

Melander

(2013) Introducing the UCDP georeferenced event dataset. Journal of Peace Research 50(4): 523–532.

32.

Tollefsen

Strand

Buhaug

(2012) PRIO-GRID: a unified spatial data structure. Journal of Peace Research 49(2): 363–374.

33.

Ward

Gleditsch

(2008) Spatial regression models. London, UK: SAGE Publications.

34.

Weiss

Traore

Kayentao

, et al. (2010) The plasmodium falciparum-specific human memory b cell compartment expands gradually with repeated malaria infections. PLOS Pathogens 6(5): 1–13.

35.

Worrall

Basu

Hanson

(2005) Is malaria a disease of poverty? a review of the literature. Tropical Medicine and International Health 10(10): 1047–1059.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.32 MB

Infectious disease and political violence: Evidence from malaria and civil conflicts in Sub-Saharan Africa

Abstract

Keywords

Introduction

Malaria and civil conflicts in Sub-Sahara Africa

Data

Empirical models and results

South Sudan as an illustrative case

Conclusion and discussion

Supplemental Material

Supplemental Material - Infectious disease and political violence: Evidence From malaria and civil conflicts in sub-saharan Africa

Footnotes

Declaration of conflicting interests

Funding

Correction (June 2025):

ORCID iD

Supplemental Material

Notes

References

Supplementary Material